the list of typed individuals. Because some of the removed parents are themselves offspring in typed trios, 2718 offspring were left. We phased these 2718 offspring/probands by (1) applying LRP to the reduced list of 31,702 individuals, and compared the results to (2) phasing the offspring using data from their parents only. From (1), 200 (7.4%) individuals could not be phased. Among the others, 2299 were phased for all SNPs, and 219 were partially phased (yield = 84.5%). The overall yield including the unphased individuals was 91.4%. Using (2), all probands were partially phased with a yield of 80.6%. Among the 978,802 heterozygous genotypes phased by both methods, there were 845 (0.086%) discrepancies (Supplementary Table 1a online). Individually, there were no discrepancy between LRP and trio phasing for 2456 (97.5%) of the 2518 offspring phased by LRP. Among the 62 probands with discrepancies, 43 had a discrepancy for only a single SNP. Considering that nearly one million phased genotypes were compared, many of these discrepancies could be attributed to miscalled genotypes in the parents, i.e. the LRP result could often be correct. Ten offspring had more than 3 discrepancies. Three of these 10 are siblings and they account for over
